An experimental evaluation of different amounts of receptive and exploratory learning in a tutoring system

摘要

A general model of knowledge acquisition is presented which distinguishes receptive and exploratory learning as two separate learning modes. In addition, learning materials are differentiated into situational examples and text information. A learner's prior knowledge is represented as heuristic rules and domain specific knowledge. The different learning methods, which may be arbitrarily combined, are implemented as PROLOG meta-interpreters so that the costs and benefits of the different learning methods can be assessed. For the first hour of learning the programming language LISP, a tutoring system was developed in which the amount of exploratory and receptive learning can be manipulated. In an experimental study with three different learning conditions, learning in a basic exploration environment (with no tutor) was compared to learning with a selective tutor and learning with a constant tutor. The results showed that the selective tutor condition was most effective: In this condition the students required the least amount of time for acquiring knowledge and solving the criterion test tasks, while solving an equal number of programming tasks correctly. The results thus show that the proposed tutoring strategy was quite successful in combining the advantages of learning by exploration and learning from instructions.